Timing, in startups, is often everything. But sometimes it's beside the point.
Intuned, a Kirkland, Washington-based company, emerged from stealth mode on June 9 with a launch post on Hacker News—a rite of passage for developer-focused startups. What made the announcement noteworthy wasn't the usual fanfare or beta waitlist. It was the claim that their platform had already processed 40 million automation runs over the past year. For a company that graduated from Y Combinator nearly four years ago, this wasn't a launch so much as a coming-out party.
The product itself addresses a problem that's become oddly central to modern software development: browser automation. Intuned's pitch is deceptively simple. Developers describe what they want to scrape or automate, and the company's AI agent generates production-ready Playwright code—actual TypeScript or Python that can be versioned, modified, and deployed like any other software. Then, crucially, the agent keeps monitoring those automations and patches them when websites inevitably change.
It's the kind of infrastructure nobody thinks about until it breaks. Which is probably the point.
The Long Road from Demo Day
Faisal and Ahmad Ilaiwi, the company's founders, have a pedigree that matters in this space. Faisal spent three years as a director of engineering at UiPath, the automation software giant, after nearly seven years at Microsoft. He applied late to Y Combinator's Summer 2022 batch and was accepted—a detail founders tend to remember—and decamped to the Bay Area for the standard three-month program.
That batch was unusual. According to TechCrunch, YC had deliberately shrunk its cohort by roughly 40 percent to about 250 companies—a response to shifting market conditions as the venture capital boom started showing cracks. Demo Day happened virtually on September 7-8, 2022, with the standard deal terms: $500,000 in total investment, $125,000 converting to a fixed 7 percent stake, and the rest on an uncapped MFN SAFE.
What's less standard is what happened next.
The company's YC profile shows four separate product launches over four years, each with distinctly different positioning. "The data assistant for engineering leaders," nearly four years ago. Then "the browser automation platform for developers and product teams" almost two years later. "Auth Sessions" for authenticated scrapers came next. Finally, the current version launched in recent months under the banner "Introducing the New Intuned."
That's not pivoting, exactly. It's something more iterative—testing different value propositions with different customer segments until one resonates. The company acknowledged this evolution in a blog post last year titled "Intuned's Act 3," though the details remain characteristically vague. Four years is a long time in startup terms, but perhaps not when you're building infrastructure that developers will trust with production workloads.
Code Over Black Boxes
The technical architecture reveals a specific bet about how AI should intersect with developer tooling. Intuned doesn't position itself as a general-purpose agent that can navigate any website on the fly. Instead, it uses AI to generate deterministic Playwright code—scripts designed to do the same thing every time, with explicit error handling and retry logic.
When a scraper breaks (and they all do, eventually), the platform's monitoring system detects the failure. The AI agent then analyzes what changed and rewrites the affected code. The distinction matters: the AI builds and maintains your automations, but it doesn't run them. At execution time, you're running standard Playwright code.
This choice has implications. Runtime AI agents that burn tokens on every page load get expensive quickly. Generated code, by contrast, runs as many times as needed without additional inference costs. The tradeoff, of course, is flexibility. You're building discrete automations for specific tasks rather than deploying a general web agent that can adapt on the fly.
The platform includes the features that separate demos from production tools: stealth mode to avoid detection, CAPTCHA solving, proxy rotation. AuthSessions, a feature for handling authenticated scraping, validates login sessions before each run and re-authenticates automatically if something breaks. It's the kind of unglamorous detail that matters when you're running thousands of scrapers.
Intuned runs on its own Chromium-based stealth browser when deployed to their cloud infrastructure, though developers can also run the generated code locally. The company ships SDKs for both JavaScript/TypeScript and Python, with recent releases appearing on PyPI (though the exact timing of these releases remains somewhat unclear from public records).
The platform has also added integration guides for running Browser-Use, an open-source agent framework that's gained some traction recently, and support for deploying Webwright scripts to their cloud. These additions, published in recent months alongside a technical deep-dive on bot detection systems, suggest the company is positioning itself as infrastructure for the broader automation ecosystem rather than a walled garden.
Finding Customers in Unglamorous Places
In their Hacker News thread, the founders described their customer base: govtech agencies, insurtech companies, and specialized agencies that build scrapers for clients. Some customers, they noted, are running more than 1,000 scrapers on the platform. The 40 million automation runs over the past year—representing some 20 million minutes of runtime—come from what the company describes as "a limited set of customers."
Reading between the lines, Intuned found product-market fit in sectors where automation is infrastructure rather than innovation theater. Government technology and insurance claims processing don't get much attention in the tech press, but they involve enormous amounts of data extraction from websites that weren't built with APIs. For those use cases, browser automation isn't a nice-to-have. It's how the work gets done.
The YC directory lists the team at around 15 people, while LinkedIn shows a range of 2-10—a common discrepancy that speaks to how opaque headcount can be at private companies, often due to reporting lag or differences in how contractors and part-time employees are counted. The company offers self-serve plans starting with a free tier, then Developer, Startup, and Enterprise options with pricing based on compute hours, job runs, and AI credits.
Testimonials on the homepage quote customers from GovSignals and Fiber AI, though these remain unverified by independent sources. Still, the 40 million runs figure—if accurate—suggests Intuned has moved well beyond proof-of-concept. Whether the platform can scale to handle the full complexity of production web scraping at the thousands-of-customers level remains an open question.
The Brittleness Problem
The broader market opportunity rests on a simple truth: web scraping remains remarkably fragile. Websites change without warning. Bot detection systems grow more sophisticated. Maintaining hundreds or thousands of scrapers becomes a full-time job, or several. The economics don't work unless you can automate the maintenance itself.
Intuned's wager is that AI can solve this problem without sacrificing the control developers need for systems they'll deploy in production. It's a narrower bet than "AI can browse the web like a human," and maybe more pragmatic. The challenge will be handling edge cases—the weird authentication flows, the dynamic JavaScript frameworks, the CAPTCHAs that evolve faster than detection workarounds.
Four years from Demo Day to a proper public launch isn't slow by any reasonable measure. It's deliberate. The question now is whether the deliberation produces a category-defining company or simply a tool that solves a real problem for a niche market. In developer tools, there's no shame in the latter. Sometimes infrastructure is just infrastructure, valuable precisely because it's boring.
The founders seem comfortable with that possibility. In an industry obsessed with blitzscaling and land grabs, Intuned's trajectory looks almost quaint—building something customers pay for, iterating based on what works, going public only when there's something substantial to show. Whether that patience translates into venture-scale outcomes is another matter entirely. But after 40 million automation runs, it's hard to argue they haven't found something real.
